ASCII / Baudot Article [telecom]

P.S. There are interesting pages preceeding the above mentioned article. check out the Teletype keyboard layout:

W.U. "didn't have a dog in that hunt" as the expression goes. Time sharing, when it was not served by direct terminal-to-computer connection, was served over dialup telephone connections using modems, so it was all a telephone company enterprise.

If I remember correctly IBM participated in the ASCII standardization effort but protested all along. Suddenly IBM just withdrew all its objections and the ASCII standard went through. Then IBM announced EBCDIC; and it was pretty obvious they hoped to use their market share to make EBCDIC the de-facto standard and bury ASCII.

The point of contention was the relationship between the communication code and the Hollerith punched card code. The most straightforward translation of Hollerith into a binary code leaves some gaps in the middle of the alphabet. There is a gap of seven characters between i and j, and another gap of eight characters between r and s. This comes about because a through i are mapped into 1 through 9 in Hollerith, with some "zone bits" to distinguish them from digits. And then j through r are similarly mapped into 1 through 9. EBCDIC preserves this gap, and allows for simple logic circuits to translate between Hollerith and EBCDIC. The gap is distasteful to have in a code because it upsets the "collating sequence", the sorting order of characters - you'd like to have the alphabet be contiguous with no odd characters stuck in the middle. And there were not enough characters available in seven-bit ASCII to just leave unassigned gaps in the middle of the code.

In the end it was all so trivial because a mere 256 bytes of memory could contain a table to translate between one code and the other. But IBM seemingly had reservations about using even that much memory for code conversion - their highly successful 1401 computer was produced in a model having merely 1400 seven-bit characters of memory in total.

The U.S. government had insisted that ASCII would be a requirement for future data processing equipment purchases. Therefore in System/360 IBM included an "ASCII mode" bit in the machine which did something fairly inconsequential - so much so that in System/370 they eliminated the bit.

Other computer manufacturers were rather stuck with their 6-bit character codes, and the industry standard 7-track magnetic tape. (6 bits plus vertical parity) IBM and later the others produced nine-track magnetic tape equipment; but at first there was need to read and write 7-track tapes with EBCDIC, which required some bit-stomping.

Previously on this topic it was said that W.U.'s big weakness was in that 'last mile' to the customer's premises--it did not have much of a local loop plant and depended on the Bell System. Is it safe to assume that, say a stock broker who had a ticker and some newswire Teletypes and perhaps a Telex machine, used Bell wires between the office and the nearest W.U. switching facility or line center?

But what about very large government, corporate, and industrial facilities located in cities? Didn't W.U. have a network of its own loop plant to serve such organizations? (I recall seeing lots of WUTC manholes in downtown Phila). What about all the companies that had Desk Fax and 'turn-knob' message requesters, weren't they connected by direct wire? Didn't the company ADT start out by providing local loop?

Anyway, I'm surprised W.U. didn't try to jump in and grab some of that early time sharing business through its own network. The articles in the Tech Review seemed to indicate they were well aware of computer based messaging and were developing interfaces for it, both for their own purposes and to serve customers.

Thanks.